France Time: 13:30-13:35
France Time: 13:35-14:30
Dr. Kevin Tierney, Professor, University of Bielefeld, Germany
Title: Combining Machine Learning and Optimization for Parallel Discrete Optimization on the GPU
Abstract: Performing discrete optimization on the GPU has long been a challenge, as the GPU is not designed for the types of computations present in standard exact and heuristic optimization methods. In contrast, deep neural networks can be easily queried on the GPU and can be easily parallelized within a single GPU, or across multiple GPUs. However, until recently, deep neural networks have seen little use in optimization techniques. Recent advances have enabled deep neural networks to be embedded into heuristic optimization techniques to assign the values of decision variables, thus opening the door to the parallelization of heuristic optimization techniques. This talk will describe the state-of-the-art for using deep learning to solve combinatorial optimization problems and provide a research agenda for the future.
Speaker Biosketch: Kevin Tierney is the Professor for Decision and Operation Technologies in the faculty of Business Administration and Economics at Bielefeld University in Germany. He was previously an assistant professor at Paderborn University in Germany and earned his PhD from the IT University of Copenhagen in Denmark. Recently, he won first place in both tracks of the IJCAI 2021 AI4TSP competition, won the EURO award for the best review paper in 2021, and his work on Neural Large Neighborhood Search won a distinguished paper award at ECAI 2020. His research interests include methodological research on using machine learning to better solve optimization problems, as well as applied research on solving maritime and other logistics problems.
France Time: 14:30-15:00
France Time: 15:00-17:00
France Time: 15:00-15:22
Paper Title: Parallel Beam Search for Combinatorial Optimization.
Authors: Nikolaus Frohner, Jan Gmys, Nouredine Melab, GŁnther Raidl and El-Ghazali Talbi.
France Time: 15:22-15:44
Paper Title: Towards a GPU Accelerated Selective Sparsity Multilayer Perceptron Algorithm using K-Nearest Neighbors Search.
Authors: Bruno Henrique Meyer and Wagner M. Nunan Zola.
France Time: 15:44-16:06
Paper Title: Training Reinforcement Learning Models via an Adversarial Evolutionary Algorithm.
Authors: Mark Coletti, Chathika Gunaratne, Catherine Schuman and Robert Patton.
France Time: 16:06-16:28
Paper Title: Frequency Recovery in Power Grids using High-Performance Computing.
Authors: Vishwas Rao, Anirudh Subramanyam, Michel Schanen, Youngdae Kim, Mihai Anitescu and Ignas Satkauskas.
France Time: 16:28-16:50
Paper Title: Accelerated Computation and Tracking of AC Optimal Power Flow Solutions Using GPUs.
Authors: Youngdae Kim and Kibaek Kim.
France Time: 16:52-17:00
CLOSING REMARKS AND ADJOURN